Machine Learning Research Group, Senior Researcher  MTA SZTAKI Kende u. 13-17, 1111 Budapest, Hungary Office: K 303 Email: Phone: (+361) 279-6262 Research Interests reinforcement learning games (chess, Go, Poker, LOA), search control neural networks optimisation algorithms for combinatorial problems

Publications

György, A., Kocsis, L., Szabó, I., Szepesvári,Cs. Continuous Time Associative Bandit Problems IJCAI-07, 830-835, 2007. (corrected version) [pdf]

Kocsis, L., Szepesvári, Cs. Bandit based Monte-Carlo Planning[pdf] ECML-06, LNCS/LNAI 4212, 282-293, 2006.

Kocsis, L., Szepesvári, Cs. Universal Parameter Optimisation in Games Based on SPSA [pdf] Machine Learning, Special Issue on Machine Learning and Games, 63, 249-286, 2006.

Kocsis, L., Szepesvári, Cs., Winands, M.H.M. RSPSA: Enhanced Parameter Optimisation in Games [ps] Advances in Computer Games'05, LNCS 4250, 39-56, 2006.

Kocsis, L., Szepesvári, Cs. Reduced-Variance Payoff Estimation in Adversarial Bandit Problems (extended version) [ps]

Kocsis, L., Learning Search Decisions, PhD thesis, Universiteit Maastricht, 2003. [ps]

Kocsis, L., Herik, H.J. van den, Uiterwijk, J.W.H.M., Two learning algorithms for forward pruning, ICGA Journal, 26(3), 165-181, 2003. [ps]

Kocsis, L., Uiterwijk, J.W.H.M., Postma, E.O., Herik, H.J. van den, The Neural MoveMap Heuristic in Chess, Computers and Games 2002, LNCS 2883, 154-170. [ps]

Winands, M.H.M., Kocsis, L., Uiterwijk, J.W.H.M., Herik, H.J. van den, Temporal Difference Learning and the Neural MoveMap Heuristic in the Game of Lines of Action, GAME-ON 2002, 99-103. [pdf]

Winands, M.H.M., Kocsis, L., Uiterwijk, J.W.H.M., Herik, H.J. van den, Learning in Lines of Action, BNAIC 2002, 371-378. [ps]

Winands, M.H.M., Kocsis, L., Uiterwijk, J.W.H.M., Herik, H.J. van den, Learning in Lines of Action, 7th Computer Olympiad Computer-Games Workshop, 2002.

Kocsis, L., Uiterwijk, J.W.H.M., and Herik, H.J. van den, Search-Independent Forward Pruning, BNAIC 2001, 159-166. [ps]

Kocsis, L., Uiterwijk, J.W.H.M., Learning Move Ordering in Chess, 6th Computer Olympiad Computer-Games Workshop, 2001.

Kocsis, L., Uiterwijk, J.W.H.M., Herik, H.J. van den, Move Ordering using Neural Networks, IEA/AIE 2001, LNCS 2070, 45-50. [ps]

Kocsis, L., Uiterwijk, J.W.H.M., Herik, H.J. van den, Learning Time Allocation using Neural Networks, Computers and Games 2000, LNCS 2063, 170-185. [ps]

Pop, T., Kocsis, L., Adaptive Strategies in the Game of Pente, Transactions on Automatic Control and Computer Science, vol. 44(58), 59-65, "Politehnica" University Press, Timisoara, 1999.

Kocsis, L., Szirbik, N.B., An Unsupervised Training Connectionist Network with Lateral Inhibition, IEA-AIE 1998, LNCS 1416, 603-611.

Kocsis, L., Szirbik, N.B., A Connectionist Method for Reducing Clustering Dimension, Transactions on Automatic Control and Computer Science, vol. 42(56), 185-193, "Politehnica" University Press, Timisoara, 1997.